The data generator is a CLI tool that simulates realistic sensor data from industrial assets like motors, pumps, and compressors. It supports both healthy and faulty operation modes for comprehensive testing.
Overview The generator creates synthetic sensor readings and sends them to the /api/v1/data/simple ingest endpoint. This is useful for:
Testing anomaly detection models
Calibrating baseline thresholds
Demonstrating fault scenarios
Load testing the ingestion pipeline
Basic Usage
Start the backend server
Ensure the FastAPI backend is running before generating data: uvicorn backend.api.main:app --reload
The generator will POST to http://localhost:8000/api/v1/data/simple by default.
Run the generator
Execute the script from the project root: python scripts/generate_data.py --asset_id motor_01 --duration 60 --healthy
Verify data ingestion
Check the terminal output for successful events: [OK] Sent event 3f5a8b2c... | V=230.5V, I=15.2A, PF=0.92, Vib=0.150g
Command-Line Arguments Asset Identification Unique identifier for the asset being simulated.
Default: Motor-01Example: motor_01, pump_alpha, compressor_3
python scripts/generate_data.py --asset_id pump_alpha
Duration How long to run the generator (in seconds).
Default: 60Range: Any positive integer
python scripts/generate_data.py --duration 300 # Run for 5 minutes
Interval Time between consecutive sensor readings.
Default: 1.0 (1 second)Range: Any positive float
python scripts/generate_data.py --interval 0.5 # 2 readings per second
Operating Mode Healthy Mode (Normal Operation) Generates normal operating sensor values:
Voltage: 230V ± 3V (Indian Grid standard)Current: 15A ± 1APower Factor: 0.88 - 0.95 (good efficiency)Vibration: 0.15g ± 0.02g (low vibration)
python scripts/generate_data.py --healthy --duration 120
Faulty Mode (Anomalous Operation) Injects extreme anomalies to force CRITICAL risk states. Fault types include:
Voltage Spike 320V ± 20V (+90V from normal)
Vibration Drift 3.0g ± 0.5g (20x normal levels)
Power Factor Drop 0.35 - 0.50 (severe degradation)
Catastrophic Failure All signals fail simultaneously
python scripts/generate_data.py --faulty --duration 30
The --healthy flag overrides --faulty if both are specified. This ensures explicit control over the operation mode.
Payload Structure Each event is sent as a JSON payload to the /api/v1/data/simple endpoint:
{ "asset_id" : "motor_01" , "voltage_v" : 230.5 , "current_a" : 15.2 , "power_factor" : 0.92 , "vibration_g" : 0.1523 , "is_faulty" : false }
Console Output Successful events display real-time metrics:
============================================================ PREDICTIVE MAINTENANCE - DATA GENERATOR ============================================================ Asset ID: motor_01 Duration: 60 seconds Interval: 1.0 seconds Mode: HEALTHY (normal operation) Endpoint: http://localhost:8000/api/v1/data/simple ============================================================ [OK] Sent event 3f5a8b2c... | V=230.5V, I=15.2A, PF=0.92, Vib=0.150g [OK] Sent event 7a9c4d1e... | V=229.8V, I=14.9A, PF=0.91, Vib=0.148g [OK] Sent event 2b8f3a5d... | V=231.2V, I=15.4A, PF=0.93, Vib=0.152g ============================================================ [COMPLETE] Sent 60 events, 0 failed ============================================================
Faulty events are marked with [FAULT]:
[OK] Sent event 4e7c9a2b... | V=315.2V, I=15.1A, PF=0.91, Vib=2.850g [FAULT]
Common Recipes Initial System Calibration Generate 5 minutes of healthy data for baseline training:
python scripts/generate_data.py \ --asset_id motor_01 \ --duration 300 \ --interval 1.0 \ --healthy
Fault Scenario Testing Simulate a brief fault condition:
python scripts/generate_data.py \ --asset_id motor_01 \ --duration 10 \ --interval 0.5 \ --faulty
High-Frequency Monitoring Generate data at 10Hz (100ms intervals) for stress testing:
python scripts/generate_data.py \ --asset_id motor_01 \ --duration 60 \ --interval 0.1 \ --healthy
Multi-Asset Simulation Run multiple generators in parallel (separate terminals):
# Terminal 1 - Motor python scripts/generate_data.py --asset_id motor_01 --healthy # Terminal 2 - Pump python scripts/generate_data.py --asset_id pump_alpha --healthy # Terminal 3 - Compressor python scripts/generate_data.py --asset_id compressor_3 --faulty
Environment Configuration Custom API Endpoint Override the default API URL using an environment variable:
Linux/Mac
Windows (PowerShell)
Windows (CMD)
export API_URL = https :// predictive-maintenance-uhlb . onrender . com python scripts/generate_data.py --asset_id motor_01 --healthy
Connection Error Handling If the backend is not running, you’ll see:
[ERROR] Cannot connect to http://localhost:8000/api/v1/data/simple Make sure the backend is running: uvicorn backend.api.main:app
Use Ctrl+C to gracefully stop the generator. It will display a summary of events sent and failed.
Advanced Usage Programmatic Integration You can import and use the generator functions in your own scripts:
import sys import os sys.path.insert( 0 , os.path.dirname(os.path.dirname(os.path.abspath( __file__ )))) from scripts.generate_data import generate_healthy_reading, send_event # Generate a single healthy reading data = generate_healthy_reading() print (data) # {'voltage_v': 230.5, 'current_a': 15.2, 'power_factor': 0.92, 'vibration_g': 0.15} # Send to API send_event( asset_id = "motor_01" , sensor_data = data, is_faulty = False )
Troubleshooting HTTP 503 Service Unavailable Cause: Backend is starting up (Render free-tier cold start).Solution: Wait 30-60 seconds and retry. The frontend’s /ping heartbeat prevents this in production.HTTP 422 Unprocessable Entity Cause: Invalid payload schema.Solution: Ensure all required fields are present: asset_id, voltage_v, current_a, power_factor, vibration_g.Data Not Appearing in Dashboard
Check InfluxDB connection: Visit /health endpointVerify bucket name: Ensure INFLUX_BUCKET=sensor_data matches your configurationInspect API logs: Look for ingestion errors in the backend console
Next Steps
Calibration Workflow Use generated data to calibrate baseline models
API Reference Learn about the /ingest endpoint schema